The present invention relates to improving the quality of extruded annular products, particularly products produced by plastic resin extrusion lines and most particularly blown plastic film.
In making such cylindrical products, the material from which the product is formed is extruded from an annular extrusion die and pulled along the die axis. In the case of blown film, plastic resin is extruded from a heated extruder having an annular die and the molten polymer is pulled away along the die axis in the form of an expanded bubble. After the resin cools to a set diameter as a result of application of cooling air, the bubble is collapsed and passes into nip rolls for further manufacturing steps.
As the film is extruded, thickness variations occur about the circumference of the bubble. The presence of thickness variations creates problems for downstream conversion equipment such as printing presses, laminators, or bag machines. In processes where the film is not converted in-line, but is wound onto a roll prior to converting, the thicker and thinner areas of many layers on the roll create hills and valleys on the roll surface which deform the film and magnify the subsequent converting problems especially with larger diameter rolls. It is therefore desirable to minimize such thickness variations, not only in blown film but in other extruded cylindrical products as well. To achieve this goal, processors use expensive equipment designed to randomize the position of these thick and thin areas over time or to automatically reduce the magnitude of these variations so that the finished roll is suitable for later converting steps.
It is recognized that thickness variations are caused by a variety of factors such as circumferential nonuniformity in flow distribution channels (ports and spirals) within the die, melt viscosity nonuniformity, and inconsistent annular die gaps through which the polymer exits the die. Flow distribution problems inside the die are of particular concern because they typically take the form of relatively sharp, closely spaced high and low spots which are commonly referred to as xe2x80x9cport linesxe2x80x9d. Additionally, variability of the cooling air and non-uniformity of air aspirated into the cooling air stream from the atmosphere surrounding the extrusion line are major contributors to film thickness variation. Many film processors rely on conventional blown film equipment to determine the film thickness. This approach typically yields an average variation of +/xe2x88x9210 to 20% in film thickness overall, with the largest contributor typically being that of port lines.
It is desired to make improvements in the die to obtain higher quality film and other products so that the downstream equipment can be run faster and longer and so that the end use products will have more consistent thickness.
One major difficulty to overcome in designing a die is how to uniformly convert a typically non-uniform flow of molten polymer or other material that is conveyed to the die via a xe2x80x9cmeltxe2x80x9d pipe into a relatively thin annular flow. Annular flow implies that there is an inner and outer forming wall as opposed to just an outer enclosing wall such as exists with the melt pipe. To introduce this inner forming wall into the molten stream requires that this new inner forming wall be rigidly fixed within the cavity of the outer enclosing wall of the die. To do this, connecting structures must be placed within the flow path of the molten material that temporarily disrupt the flow forming multiple, separate flows which then pass by the connecting structures and must be recombined in some way. Unfortunately, molten polymer exhibits non-uniform melt viscosity due mainly to variations in molecular level properties as well as local polymer temperature. These viscosity effects are collectively referred to as the rheology. One such property of major concern is that polymers exhibit xe2x80x9cnon-Newtonianxe2x80x9d flow behavior. This means that the viscosity of the polymer changes depending on how fast it is moving through a given channel. The net effect when all viscosity effects are combined is that the polymer tends to segregate by viscosity making uniform recombination of multiple polymer flows very difficult. Additionally, molten polymer remembers its previous flow history and instead of seamlessly recombining, the multiple polymer flows tend to form unwanted xe2x80x9cweld linesxe2x80x9d where adjacent flows are recombined. The problem of weld lines intensifies when degradation of the polymer occurs due to low polymer flow rates.
Several approaches are presently employed to provide for connecting structure between the outer and inner forming walls of the die. One approach feeds from the centerline axis, a small distribution chamber in the die. This chamber separates and directs the polymer into several smaller, equally spaced pipes called ports, which diverge radially at some angle to the flow axis of the incoming melt. These ports convey the polymer out to a diameter appropriate for recombining into the annular flow which will exit the die. Another approach creates a mushroom shaped distribution chamber out of which relatively small, highly streamlined, spider-like connecting structures diverge radially at an angle to the flow axis that allow for quick recombination before forming the generally axial annular flow that exits the die. Yet another approach feeds the die radially from the side of the die and divides the flow one or more times through a network of flow channels similar to the branches of a tree, which ultimately convey the separate polymer streams to a diameter appropriate for recombining into the annular flow which will exit the die. Generally, one or more of the methods of flow separation must be employed in a blown film die, but each causes problems with segregation and potential for weld lines to form. Special recombination techniques must be employed to limit these effects.
Several techniques are used to recombine individual molten material flows into the annular flow that exits from the die. Some are designed to overlap the separate flows creating an onion-like layering effect, while others simply butt opposed flows up against each other and allow time, temperature and pressure to force recombination to occur.
In blown film production, the most common recombination technique commercially available employs channels which spiral around the axis of the die. These so-called spirals overlap one another and allow molten polymer to gradually bleed out of the channel over a xe2x80x9clandxe2x80x9d, eventually to flow toward the annular exit of the die forming a layered, almost onion-like recombination flow. This annular flow of polymer exits the die at what is commonly referred to as the die lip. The major problem with this approach is that the flow channels and lands must be made non-uniform to compensate for non-Newtonian flow and other non-uniformities exhibited by the polymer. Unfortunately, major differences exist in the flow characteristics of various polymer materials that are processed. For a given die design, it may be possible to obtain even distribution around the flow annulus for one material, however it will not be even for others. Instead, other materials tend to form somewhat sinusoidal high and low flow spots in locations which depend on the material properties being processed. Thus the spiral design approach is limited in its capability to process a broad range of materials while simultaneously holding thickness variations to a consistent, predictable minimum.
A further problem is that the polymer or other material must necessarily take a long period of time to flow through the passages, i.e., a high residence time, which can lead to degradation of the material. Additionally, as the material flows through each passage, significant back pressure is created.
In xe2x80x9cpancakexe2x80x9d designs which incorporate distribution channels and the spirals substantially into the face of a plate that is coaxial with the flow axis of the die, the wetted surface area is quite large so that, when combined with higher pressures, resulting separation forces between adjacent plates can grow to be so large that the die cannot be held together. This forces the designer of such dies to limit the pressure magnitude which tends to degrade even distribution. Further, in many cases, lower pressure is attained by enlarging the flow passages; however this leads to higher residence time causing degradation of polymer properties. In practice, pressure and distribution effectiveness must be balanced which can lead to limitations on how large the die can be.
A less commonly used recombination approach does not overlap the flows but instead joins them at one or more discrete locations. In these locations where two opposed flows join together, the flow is very low causing the material to have very long residence times which degrades the polymer. This degraded polymer forms a distinct weld line that exhibits poor optical properties and reduced strength which have tended to limit the use of these designs. On the other hand, since there is no overlap, the flow channels are shorter than in overlap designs. This provides benefits in lower pressure and residence time which limits degradation and allows for larger designs. Non-overlapping designs also benefit from the clearly defined flow paths which force the polymer through the same geometry regardless of melt flow characteristics as opposed to the shifting around of the flow path associated with overlapping designs. This simplifies the die design process since non-Newtonian flow is well understood through defined geometries. Unfortunately, non-uniformities in distribution still occur as the melt flow characteristics change from those that were used to design the die. As a wider range of polymer choices are made available, this becomes more of a problem.
Processors are presented with a growing number of choices of extrusion materials, each with their own special properties. For example, some polymers resist water vapor, others resist oxygen penetration, still others provide high strength or resist puncture. Increasingly, processors are finding innovative uses for these materials, oftentimes finding it desirable to combine different polymers together in a layered or xe2x80x9ccoextrudedxe2x80x9d structure to yield property benefits in several areas. To do this, dies are designed with multiple entry points which distribute the polymer flow into separate annular flows and subsequently layer these flows one inside the other while still inside the die. Although non-overlapping designs have been used, most prevalent are overlapping designs either in a concentric or pancake configuration. Pancake designs are better suited to larger numbers of layers because the individual layers can be stacked one on top of each other. Concentric designs are limited to about 5 to 7 layers simply because the die grows so large in diameter as to become impracticable.
It has long been recognized that having multiple layers can provide a secondary benefit in that thickness variations present in each layer can somewhat offset one another. This has a drawback; since each layer""s variation depends on associated melt flow properties, throughput rate, temperature, etc., the variations typically will not always average out. In fact, they can even align one on top of each other yielding no thickness averaging whatever. This is especially true of overlapping designs since the melt variations shift significantly in position and magnitude with even subtle changes in a given layer. Commercial coextrusion dies are designed with adjacent layer spirals that typically wrap in opposed directions in an effort to capitalize on this averaging effect. In the case of concentric die designs, the spirals for each layer are necessarily different in design because they do not spiral around at the same distance from the flow axis of the die. Pancake designs can be designed with the same mechanical geometry, however the path length to the die lip is necessarily different for each layer because they are stacked one on top of each other. This causes differences in the flow behavior since each layer operates at a different pressure. It has been observed that commercially available dies designed to capitalize on averaging effects exhibit both very good and very bad variation in total thickness as the throughput rate is raised through its full operating range. This occurs as resultant layer variations first oppose (good) then align (bad) with one another. An additional problem with these designs is that even if thickness variations are opposed, yielding good overall variation, the individual layer distribution can still be bad. This has a negative effect, especially when each layer is designed to take advantage of different film propertiesxe2x80x94the layers responsible for providing a barrier to oxygen and separately to water vapor can individually be highly variable even though the total thickness is uniform. It is highly desirable to achieve uniform distribution for each individual layer as well as for the combination of multiple layers.
The present invention features a regular division (RD) die which provides uniform distribution of molten extrusion material to each individual layer and exhibits a high degree of insensitivity to melt flow properties and a pressure resistive distribution system that does not limit the size of the die. This die design has particular application to the extrusion of polymeric blown film, but also applies to other forms of extrusion requiring an annular die. Blown film extrusion lines typically include a heated extruder for melting and pressurizing a flow of molten plastic resin, an annular die through which the molten resin extrudes and from which it is pulled away along an axis in the form of an expanding bubble, and an air cooling device constructed to direct cooling air into cooling contact with the bubble, to flow along the bubble and cause the molten resin to cool as the film expands until a substantially fixed maximum bubble diameter is achieved at a frost line spaced from the annular die.
The RD design may be included as an integral part of one or more individual die layers within the complete die. According to one preferred embodiment, the RD design is integrated separately in each layer of a pancake style stackable die. Each layer includes a series of concentric rings one inside of the other that performs the functions of feeding, distribution, and recombination. These rings surround and contact one another to allow the polymer to pass between them unimpeded through passages cut into the surfaces of and/or through them. The rings are bolted together forming a single unitized layer that is stacked face to face with the other layers of the complete die, each layer with its central geometrical axis being coaxial with the flow axis of the die. Polymer is separately fed into the outside diameter of the outer feed ring of each layer, the polymer passing straight radially through the feed ring wall to the radially interior associated distribution ring. For purposes of the ensuing discussion, the location of the input through the feed ring is at location 0xc2x0.
The distribution ring has flow channels machined into its radially outwardly-facing surface which act to divide the flow one or more times. Cutting the channels into the outside surface (or alternatively, the radially inwardly-facing surface, or both) eliminates the detrimental effects of separation forces caused by polymer pressure; the forces produced by the polymer act against the surrounding feed ring instead of on the bolts which hold the layer(s) together.
In the distribution ring, the polymer flow input from the feed ring is divided into an even number (2n) of separate and equal flows. In the preferred embodiment, the input flow is divided into eight (23) flows, in three stages. The first division of flow occurs at 0xc2x0, at which point the polymer flow is divided in two and each half is directed into one of two channels, each of which wraps 90 degrees around the circumference of the ring, one clockwise from 0 degrees to 90 degrees and the other counter-clockwise from 0 degrees to 270 degrees. At the 90 and 270 degree points, each flow (half of the original) turns and travels axially for a short distance prior to being divided a second time. The second divisions occur separately at the 90xc2x0 and 270xc2x0 points; at each of which the flow is again divided in half and the resulting portion of the flow (one quarter of the total input flow) directed into one of a pair of channels which wrap 45xc2x0 in opposite directions from, respectively, the 90xc2x0 and 270xc2x0 points, around the outside of the ring. These four flows end up at, 45xc2x0, 135xc2x0, 225xc2x0 and 315xc2x0; at which points the flow is divided again, this time into opposite wrap angles of 22.5. The end result of these three divisions is eight separate flows which end at 45 degree intervals at, respectively, xe2x80x9c22.5xc2x0, 67.5xc2x0, 112.5xc2x0, . . . , 337.5xc2x0. It will be noted that, after each division, equal opposite wrap angles ensure that there is equal path length and thus equal pressure drop for any path through which the polymer might flow.
Each of these eight divided flows then passes radially inwardly through the first distribution ring, either directly to the recombination rings or, if further division is desired, to a second distribution ring. It will be recognized that, by using more than one distribution ring, a larger number xe2x80x9cnxe2x80x9d of divisions can be accomplished without pressure penalties. In any event, after the desired number of divisions are made in the distribution rings, the resulting flows are conveyed radially inwardly to the recombination rings through a divider plate that forms an integral part of the final (e.g., the most radially inward) distribution ring.
The divider plate is relatively thin (measured axially of the die) compared to the main body of the distribution ring of which it is a part. The divider plate extends inwardly from the portion of the final radially inward distribution ring that forms the 2n polymer flows and tapers to a thin edge at its inner circumference. Within the divider plate, and generally prior to the taper, the 2n radial flows are alternately diverted to one side of the plate or the other. This provides two separate but identical flow patterns, each of which includes 2(nxe2x88x921) recombination flows, issuing from ports located in either the upper or the lower face of the divider plate. These flows in turn are fed to a pair of recombination plates that abut the upper and lower faces of the divider plate.
One recombination plate is mounted on either side of the tapered portion of the divider plate. The recombination flow ports on one side of the divider plate are offset in such a way as to be centered between ports on the opposite side of the divider plate. This allows for precise, mirror image recombination to take place, xe2x80x9csplitxe2x80x9d on opposite sides of the divider plate. These split, mirror-imaged flows join together at the inner edge of the divider plate. The recombination flow channels on each side of the divider plate are designed to create a flow distribution that, when added to its mirror image, results in a flat flow profile.
Insensitivity to melt rheology is attained by forcing the recombination plate flow to distribute in a non-overlapping manner, thus yielding predictable, non-shifting resultant polymer flow. Weld lines are avoided by placing an interceding land area directly in front of each port with the main flow channel passing on a diameter behind the land. Thus some of the flow from each port passes over the land and, of what remains, half flows down the channel one way and the other half flows in the opposite direction. Eventually the channel flow from one port meets opposite direction flow from the adjacent port. At this point, the main flow channel passes radially inward between the ends of adjacent lands. This creates a weld area, but because the weld area is in a high flow region the problem of polymer degradation is substantially eliminated. The main flow channel then splits again and passes on a diameter in front of each of the associated lands such that half flows down the channel one way and the other half goes the opposite direction. Thus the flow which originally was diverted around the land via the main flow channels is recombined with the land flow in a way which is predictably stable but yields a layered effect, similar to that produced in a spiral design but without shifts in position. The now annular and radially inwardly directed recombination flow passes over a final land to the tip of the divider plate where its mirror imaged split flow from the opposite side of the divider plate is added. The final channel and land are cut in such a way as to insure a smaller flow where the high flow weld line occurs and a larger flow centered on the interceding land. Upon addition of its mirror image, the deleterious effects of the weld area is minimized by the addition of the mirror images larger (non-weld) flow area.
The shape of the flow issuing from the recombination area on each side of the divider plate prior to the flows being recombined is important to achieving a combined uniform flow from opposite sides of the divider plate. Although for a given material, the individual flows from each halve may also be uniform, they do not necessarily have to be. Rather, there is a wide diversity of curves which can be programmed into the design of the flow channels which after addition yield a uniformly flat combined profile. The mathematical study of xe2x80x9cregular divisions of the planexe2x80x9d such as used in the study of crystallography or as can be found in graphical representations by M. C. Escher depict many suitable examples of both simple and complex profiles. A preferred profile for each split flow, is a straight line xe2x80x9ctrianglexe2x80x9d profile which linearly increases from a minimum flow at the high flow weld to a maximum in line with the port. This profile repeats itself without discontinuity around the diameter of the layer. A second preferred split flow profile is a xe2x80x9csinusoidalxe2x80x9d profile which also has its minimum at the high flow weld and maximum in line with the port. The flow profiles are periodic and identical but offset 180xc2x0 in period from each other, as shown in FIGS. 6 and 6a.